Organic Transistor–Based Chemical Sensors for Real‐Sample Analysis

An organic field‐effect transistor (OFET) is the representative amplification device showing a switching profile by applying a gate voltage, which indicates the potential as a chemical sensor device in combination with appropriate molecular‐recognition materials. In contrast, the realization of OFET‐based chemical sensors for real‐sample analysis is limited owing to the instability of organic semiconductive materials under ambient conditions and the difficulty of the designs of molecular‐recognition materials. Methodologies and actual approaches to maximize the potential of the OFETs as chemical sensor platforms based on fusion technologies between organic electronics and molecular‐recognition chemistry are described in this review.


Introduction
Sensors of our human bodies play important roles to detect invisible physical and chemical stimuli, which change (or control) our behaviors nonconsciously or consciously. [1]epresentative sensors incorporated in human bodies are classified into five: eyes for lights, ears for sounds, skin for temperatures and pressures, nose for smells, and tongue for tastes.Inspired by such sophisticated sensors endowed into our bodies, the development of various sensor platforms has promoted to detection of physical and chemical stimuli "instead of ourselves."The requirements of sensor abilities are favorable sensitivity, selectivity, quick response, wide response ranges, reusability, repeatability, suppression ability of interference effects, etc., which have been desired to be "beyond the inherent recognition abilities of biological sensing systems."Sensors consist of receptors to detect invisible stimuli, and transducers (or reporters) to amplify the sensor signals. [2]hysical sensors have already been well established, while a few biosensors to evaluate specific biomarkers have been used in real-world scenarios.However, the number of detectable analytes by the biosensors is limited in practical analytical situations that require various analytes contained in human samples, environmental water, foods, drinks, etc., which is caused by the limitation of types of enzymes and antibodies.Meanwhile, supramolecular receptors can be designed based on molecular-recognition chemistry, whereas a stagnant state of development of chemical sensors is attributed to the difficulty of receptor designs considering molecular geometries and sizes, and molecular interactions with analytes. [3,4]In addition, the selection of transducers classified into optical, electrochemical, electric, mass sensitive, magnetic, thermometric, etc., significantly depends on the types and sizes of sensor platforms, and detection principles. [2]herefore, the feasibility of portable sensors for on-site detection relies on the selection of transducers.In the development of chemical sensors, not only appropriate designs of receptors and transducers but also the establishment of methodology for assembly of the receptor and transducer are required.
A field-effect transistor (FET) is a representative electronic device consisting of a gate, a source, a drain, a dielectric, and a semiconductive layer. [5,6]FETs show nonlinear switching characteristics by applying a gate voltage.[9] In the device structures for chemical sensing, the organic semiconductive layer [10] could be used as the detection portion, whereas the vulnerability of OFETs derived from the instability of organic semiconductive materials against water could bottleneck for chemical sensing. [11][14] With appropriate molecular-recognition materials, the extended-gate-type OFET can be used as chemical sensors for real-sample analysis.
This review overall focuses on the extended-gate-type OFETs from the perspective of comprehensive designs of detection portions constructed by biological and artificial materials, and their sensing applications.In the outline, Chapter 2 summarizes the basic structure of extended-gate-type OFET and the detection mechanism on the extended-gate electrode functionalized with molecular-recognition materials.Chapter 3 describes approaches to the immobilization of biological materials on the extended-gate electrodes to obtain uniform and reproducible detection portions.
In Chapter 4, approaches using cross-reactivities of supramolecular receptors for pattern recognition are summarized.Chapter 5 explains methodologies for selective detection based on the optimization of recognition sites considering molecular geometries between molecularly imprinted polymers (MIPs) and analytes.

General Introduction of OFET-Based Sensors
The basic chemical sensor design based on the extendedgate-type OFET is shown in Figure 1a.The surface changes of the extended-gate electrode upon analyte sensing can be read out by the drive unit (i.e., the OFET). [13,14]The representative commercial FET devices are metal-oxide-semiconductor FETs (MOSFETs) [15] and junction FETs. [16]The operation principle of the OFET in this review is followed by that of MOSFETs. [17]In chemical sensing using FETs, uniform and reproducible transistor characteristics and low voltage operation are required to read out analyte-capture information.The polarization of electric charges in the gate dielectric layer is induced by applying gate voltage (V GS ), and subsequently flows the charge carrier (electron or hole) in the organic semiconductive layer between the source and drain electrodes by applying the drain voltage (V DS ).One of the transistor characteristics of the OFET to evaluate device performances is described using the following equation (Equation ( 1)) W and L indicate the channel width and length of the OFET, respectively; μ represents the field-effect mobility; and C is the capacitance of the gate dielectric.The threshold voltage (V TH ) is a gate voltage to form conducting channel between the source and drain electrodes, which is one of the factors to evaluate transistor characteristics. [18]In this regard, a relationship between V TH and the charge density (Q ) is described as the following equation (Equation ( 2)) where a change in the V TH (i.e., ΔV TH ) is induced by a change in Q (i.e., ΔQ ) (Figure 1b).Therefore, V GS is a trigger to regulate the conductance of the organic semiconductive layer, indicating that the gate electrode could be employed as a sensing portion.In accordance with Equation (1), the drain current (I DS ) and the V TH of the OFET are variables in chemical sensing.In other words, both factors ΔI DS and ΔV TH can be used as quantitative sensor responses derived from analyte detection on the gate electrode functionalized with molecular-recognition materials. [19]n the extended-gate-type OFET-based chemical sensor, the V GS of the OFET is applied through a reference electrode at a certain voltage.Therefore, the changes in transistor characteristics (i.e., transfer and output characteristics) are induced by analyte detection at the interface of the extended-gate electrode and aqueous media, which is attributed to the potential difference between the extended-gate electrode and the reference electrode.The representative operation units in extended-gate-type OFETs introduced in this review are bottom-gate-bottomcontact-type OFETs.The solution-processable materials such as 3,9-dihexyldinaphtho[2,3-b:2,3-d]thiophene (C6-DNT-VW [20] ), poly{2,5-bis(3-tetradecylthiophen-2-yl)thieno [3,2-b]thiophene} (PBTTT-C14 [21] ), and poly{2,5-bis(3-hexadecylthiophen-2-yl) thieno [3,2-b]thiophene} (PBTTT-C16 [21] ) were applied to organic semiconductive layers, which enabled the application of high-throughput fabrication methods such as dispensers.In addition, double dielectric layers consisting of metal oxides and self-assembled monolayers were employed to obtain high capacitance. [22]The surface of the organic semiconductive layers was fully coated with a hydrophobic material for stable operation under ambient conditions. [14]The extended-gate electrode made of gold (Au) was prepared on a polyethylene naphthalate film substrate, whereby molecular-recognition materials could be immobilized on the surface of Au electrodes through thiols (and dithiols [23] ), alkynes, [24][25][26] etc., to obtain highly ordered recognition portions.
The modification of molecular-recognition materials on the extended-gate electrodes is characterized using various methods.In addition, multilateral evaluation of sensor responses by analyte detection is also required.Therefore, several characterization methods for electrodes are applied to examine the extended-gate electrodes before and after analyte capture.For example, the properties of extended-gate electrodes including work functions and wettability are significantly affected by the treatment of the electrode surface.Photoelectron yield spectroscopy (PYS) is a method to estimate a work function of an electrode.Therefore, the changes in the work function of the electrode indicate the immobilization of molecular-recognition materials, which are determined by electron-donating/withdrawing groups of the materials immobilized on the electrode surface.In addition, PYS can be also applied to investigate changes in dipole moments of the molecular-recognition materials on the electrodes upon analyte detection. [27]The hydrophilicity (or hydrophobicity) of the surface of the sensing electrode can be examined by wettability tests.Fourier-transform infrared spectroscopy with variable angle attenuated total reflection is employed to identify functional groups of membranes on the electrodes.Elemental analysis can be performed by X-ray photoelectron spectroscopy measurements, which enable the prediction of chemical structural changes of self-assembled monolayers (SAMs) based on peak shifts and the changes in relative atomic concentration ratios upon analyte detection.In contrast to the previous qualitative characterization methods, electrochemical methods including linear sweep voltammetry allow the estimation of the molecular density of thiolated SAMs on electrodes.The morphology of the extended-gate electrodes is examined by scanning electron microscopy, [28] atomic force microscopy, [29,30] etc.Comprehensive characterization is important to obtain reproducible sensing electrodes, as shown above.
Especially, this review introduces the methodologies to obtain uniform sensing electrodes functionalized with biological and artificial recognition materials through chemical bonds in Chapters 3 and 4 and electrochemical deposition in Chapter 5 (Table 1).

OFET-Based Sensors Functionalized with Biological Materials for Selective Detection
Real samples such as human saliva, urine, blood, etc., contain valuable and abundant biomarkers to examine normal or abnormal states of psychological and physical conditions. [31]onventionally, large-sized instrumental methods including high-performance liquid chromatography have been used for the analysis of real samples to identify components and their concentrations, whereas such reliable techniques require complicated sample treatments and procedures under the supervision of a professional person.Meanwhile, the chemical sensors enable analysis without any pretreatment process, which simplifies the sensing protocol in practical analysis.As representative biological-recognition materials, enzymes and antibodies have been widely used in biosensing owing to their favorable detection abilities against a specific analyte contained in real samples. [32]In the design of OFET-based biosensors, the establishment of immobilization methods for the biogenic materials is important to achieve accurate sensing.Therefore, this chapter describes actual approaches to obtain high reproducible sensing portion constructed by an SAM-based scaffolds on the extended-gate electrodes.

Selective Detection using Enzymatic OFET-Based Sensor
Enzymatic catalysis manners based on the principle of a lockand-key model allow selective recognition of a specific biomarker in biological-recognition systems. [4]Indeed, such inherent specificity of enzymes against a substance (i.e., analytes) has been numerously applied to selective detection of a specific analyte in the presence of excess amounts of interferents. [32]In biosensing based on enzymatic catalysis, an electron relay between the enzyme layer and the electrodes induced by adding analytes significantly affects quantitative changes in the surface potential of the electrodes.Therefore, the appropriate modification of enzyme layers considering the direction of the electron relay is necessary to effectively read out sensor responses by the enzymatic catalysis reaction.Herein, this chapter focuses on designs of enzyme-mediator-linked SAMs to achieve selective detection of biomarkers in human samples.
The first example in this section is an enzyme-mediator-linked SAM for the selective detection of a nitrate ion (NO 3 À ) in human saliva. [33]The target NO 3 À contained in foods, drinking and environmental water could be a causative agent of significant diseases including bladder, gastric cancer, infant methemoglobinemia, etc. [34,35] Moreover, given the fact that the psychological association with NO 3 À levels in human saliva was reported, [36] the quantitative detection of the salivary NO 3 À is important to visualize the relationship between psychological stress and levels of salivary NO 3 À . [34,35]In this demonstration, an enzymatic layer made of nitrate reductase was constructed on an SAM-linked bipyridinium (BP) derivative as an electron-transfer mediator (Figure 2a).By the activation of the nitrate reductase in the presence of sodium dithionite (Na 2 S 2 O 4 ), the changes in the valence of the mediator are induced (i.e., BP 2þ ↔BP þ ). [37]Therefore, the OFET showed changes in transistor characteristics originating from the electron relay on the extended-gate electrode upon adding NO 3 À in the presence of Na 2 S 2 O 4 .Certainly, the extendedgate-type OFET functionalized with the nitrate reductase-linked mediator selectively detected NO 3 À over chloride (Cl À ), thiocyanate (SCN À ), hydrogen phosphate (HPO 4 2À ), and bicarbonate (HCO 3 À ) (Figure 2b).To investigate the applicability of the manufactured OFET-based enzymatic sensor for salivary analysis, a spike-and-recovery test using a diluted saliva sample was performed.As shown in Figure 2c, OFET responses corresponding to the NO 3 À levels in the saliva sample were observed.The recovery rate was estimated to be 97.4% AE 1.8% (n = 3, addition amount: 24 μM), which was comparable to a commercially available colorimetric method (100.4% AE 5.2%).
The proposed strategy using SAM-linked mediators was further applied to urinalysis for the evaluation of the potential of the extended-gate-type OFET-based biosensor. [38]The target catecholamines (i.e., adrenaline, noradrenaline, and dopamine) contained in human urine are the representative neurotransmitters, which play essential roles in the sympathoadrenal system. [39,40]Since the fluctuating activity of the catecholamines is associated with mental disorders, psychological stress could be caused by the changes in those levels. [41]Moreover, catecholamine levels as indicators for urinalysis have been examined because excess amounts of catecholamines are produced in the abnormal states by the tumors causing significant diseases such as pheochromocytomas and paragangliomas.To monitor urinary dopamine levels using a laccase, an N-ethylphenazonium moiety-linked SAM was immobilized on the extended-gate electrode.The laccase induces the oxidation of dopamine to dopamine-1,2-quinone, followed by the reduction of the mediator layer made of the N-ethylphenazonium moiety. [41,42]Hence, quantitative shifts in transistor characteristics based on a dopamine oxidation-triggered electron relay were observed upon an increase in dopamine concentration (Figure 3a).The fabricated  OFET-based biosensor exhibited continuous changes in I DS with the increase in dopamine concentration, which stemmed from the enzymatic catalysis of laccase on the SAM-linked mediator layer.In this regard, the 80% sensor response of the OFET-based enzymatic sensor reached 70 s, suggesting the potential of the OFET-based enzymatic sensor for a rapid urinary test (Figure 3b).Next, the selectivity of the enzymatic sensor was evaluated against electrochemical interferents for urinalysis (i.e., ascorbic acid, creatinine, uric acid, and phenylethylamine).
creatinine, and phenylethylamine were observed probably due to the physical adsorption on the surface of the sensing electrode, interference effects were overall suppressed by the selection of N-ethylphenazonium moiety as the mediator allowing a large signal-to-noise ratio owing to its low redox potential (i.e., near 0.0 V versus Ag/AgCl). [41,42]Furthermore, the feasibility of the OFET-based enzymatic sensor for urinalysis was examined by a spike-and-recovery test against a non-diluted human urine sample without any pretreatment (Figure 3d). Figure 3d represents a calibration line obtained by the collection of transistor responses (i.e., V GS -I DS ) to dopamine at different concentrations.The determined recovery rates of four plots distributed on the calibration line were 97%-104% at 2.55-3.75μM in the human urine sample, and the accuracy was comparable to that of a medical instrument.

Selective Detection using OFET-Based Immunoassay
Immunoassay is a well-established analytical method for selective detection based on the specificity of antibodies. [32]Indeed, the applicability of antibodies for real-sample analysis has been widely investigated using various sensor platforms, whereas the reproducible analysis based on immunoassay has been a long-standing issue because of the difficulty of the immobilization of antibodies considering their orientations. [43]Especially, the orientation of antibodies and analytes such as proteins and hormones that possess multiple charges and complicated structures significantly influences sensor responses in solid-state sensing fashions.For example, analytes possessing multiple charges are spontaneously and randomly adsorbed on the surface of sensing electrodes without any surface treatment (Figure 4a). [44]In this case, the disorder of charge distribution could cause fluctuation of sensor responses.Meanwhile, the charged analytes are also captured on an electrode directly immobilized with antibodies, whereas the contribution of antibodies for recognition is not significant because the recognition surfaces of the antibodies are not aligned on the electrode (Figure 4b). [45]In contrast, the SAM-based scaffolds (e.g., biotin-streptavidin complexes) are capable of forming uniform surfaces to immobilize largesized antibodies with high orientation (Figure 4c). [43]Hence, the antibodies linked with SAMs enable to capture of multivalent charged analytes such as proteins and hormones effectively through multiple noncovalent interactions.Indeed, the approach has shown a favorable effect on quantitative analyte detection using OFET-based chemical sensors. [43]he approach to control the orientation of recognition scaffolds on the electrode was applied to the selective detection of a hormone showing a complicated chemical structure in human saliva using an extended-gate-type OFET. [46]The target oxytocin comprising nine amino acid residues plays an important role in biological systems as a neurotransmitter. [47]Significant roles of oxytocin including an age-specific circulating hormone, [48] an anabolic bone hormone, [49] and medicine for chronic abdominal pain in peripheral tissue [50] have been reported, indicating that oxytocin is an essential biomarker.Moreover, peripheral effects of oxytocin associated with psychological perspective have also been researched, whereas the details of mechanisms in biological systems have not been cleared. [51,52]Furthermore, quantitative detection of salivary oxytocin would contribute to monitoring the relationship between peripheral effects and psychological changes, [53] which could accelerate fundamental research in the fields of molecular biology, pharmacology, maternity nursing, and diagnosis. [51,52]In this immunoassay, an antibodyattached SAM was selected to achieve ultrasensitive and selective detection of oxytocin in human saliva.A short alkyl chain-based SAM was employed considering chemical sensing within the Debye length, which allows highly sensitive detection. [54]In addition, a biotinylated anti-oxytocin antibody was immobilized on a streptavidin-attached SAM through biotin-avidin interactions (Figure 5a).The OFET connected to the extended-gate electrode functionalized with the antibody-linked SAM showed the highest response to oxytocin over chemical species contained in human saliva (i.e., glucose, creatinine, lactic acid (LA), uric acid, potassium ion (K þ ), calcium ion (Ca 2þ ), and vasopressin) (Figure 5b).Notably, vasopressin showing a similar structure in comparison to oxytocin was distinguished from response to oxytocin, which suggested the discriminant power of the OFET-based immunosensor.Indeed, the manufactured OFET-based immunosensor exhibited stepwise shifts of transistor characteristics upon adding oxytocin concentration in human saliva, and the limit of detection (LoD) [55] was estimated to be 3.9 pg mL À1 indicating highly sensitive detection.The OFET-based immunosensor was further applied to a spike-and-recovery test to evaluate the prediction ability against salivary oxytocin without any pretreatment.As shown in Figure 5c, the predicted datasets (red circles) were accurately distributed on a built calibration line (black squares), resulting in recovery rates of 96% and 102% at 20 and 25 pg mL À1 .Considering the actual oxytocin levels in the case of pregnant and lactating women's saliva (10-40 pg mL À1 ), [56] the potential of the OFET-based immunosensor for practical diagnostics was revealed by the real-sample analysis.
As shown earlier, selective detection was achieved by the inherent detectability of biogenic materials based on lock-andkey models, while the libraries of enzymes and antibodies are limited.Thus, approaches to design detection portions based on molecular-recognition chemistry are required for sensing various analytes.

OFET-Based Sensors Functionalized with Supramolecular Receptors for Pattern Recognition
As shown in Section 3, the extended-gate-type OFET functionalized with biological materials has achieved selective and sensitive detection of biomarkers in human samples.In contrast, the instability of biogenic materials could limit sensing environments in practical analysis.Hence, the employment of artificial materials is one of the appropriate approaches toward chemical sensing in real-world scenarios, whereas the cross-reactivity of supramolecular receptors based on the principle of molecular-recognition chemistry could be a bottleneck in selective detection for a specific analyte. [3,4]Meanwhile, such inherent cross-reactivity can be used for the simultaneous discrimination of various analytes combined with pattern-recognition techniques. [57,58]The concept of pattern recognition using sensor arrays is attributed to the mammalian olfactory system allowing the discrimination of various odorant molecules. [59]][62] The inherent cross-reactivity derived from supramolecular receptors contributes to obtaining an information-rich dataset referred to as a fingerprint-like response for pattern recognition. [63]ndeed, organic conductive materials functionalized with supramolecular receptors have been used as platforms for electronic noses to detect multiple volatile organic compounds (VOCs) in combination with computational methods. [64,65]For example, Swager's group designed a conductive chemiresistor array consisting of multiwalled carbon nanotube functionalized with different supramolecular receptors possessing hexafluoroisopropanol and carboxylic acid groups (for ethers and ketones), amide-linked crown-ether groups (for acids and alcohols), acetylenedicarboxylate esters (for vapors with high polarity such as ketones and ethers), calix [4]arenes (for aromatic and chlorinated hydrocarbons), and long-chain dodecyl groups (for aliphatic hydrocarbons) (Figure 6a). [65]The sensor array showed various ), creatinine (0.3 μg mL À1 ), LA (18 μg mL À1 ), uric acid (8 μg mL À1 ), K þ (547 μg mL À1 ), Ca 2þ (48 μg mL À1 ), vasopressin (10 pg mL À1 ), and oxytocin (10 pg mL À1 )).c) Spike-and-recovery test for oxytocin in the human saliva sample.Reproduced with permission. [46]Copyright 2022, The Royal Society of Chemistry.
conductive responses (ÀΔG/G 0 ) to 20 types of VOCs, obtaining a fingerprint-like response pattern (Figure 6b).The obtained responses were applied to principal component analysis (PCA) as an unsupervised multivariate analysis method [66] to perform pattern recognition of VOCs classified into alcohols, ethers, ketones, aliphatic, aromatic, and chlorinated hydrocarbons.As shown in the PCA score plot (Figure 6c), all clusters showing similarity of analyte structures were distributed and categorized, which was achieved by the recognition ability of the employed supramolecular receptors.
In contrast to gas sensor devices based on organic conductive materials, the development of organic devices for electronic tongues to detect multiple analytes in aqueous media has been still challenging.Thus, Minami et al. proposed an approach to introduce supramolecular receptors into the extended-gate-type OFET for the realization of electronic tongues. [67,68]In solid-state chemical sensing, supramolecular receptors can be introduced as SAMs into the chemical sensor devices.This section shows actual examples of OFET-based chemical sensors functionalized with supramolecular receptors for cross-reactive detection. [67,68]he first example of the electronic tongue in this section is the qualitative and quantitative detection of oxyanions in human blood serum using an OFET-based chemical sensor functionalized with a complex of a thiolated-dipicolylamine (dpa) derivative and a copper(II) ion (Cu II -dpa). [67]Target oxyanions are ubiquitously distributed in nature, and play crucial roles in biological systems.Especially, phosphates in human blood serum at high concentrations cause an increase in the risk of death, which leads to heart failure and myocardial infarction.[71][72] However, the simultaneous discrimination of various anions in water is still challenging because of their structural diversity. [73,74]For the recognition of the tetrahedral structure of hydrogen monophosphate (HPO 4 2À , Pi), sophisticated receptor motifs including a pseudo tetrahedral cleft, [75][76][77] a tripodal skeleton, [78,79] macrocyclic-based coordination, [78,79] and acyclic-dinuclear metal complexes [80] have been designed and synthesized. [78,79]Among them, the dpa-based metal complexes are the representative coordination receptors for anion recognition.84][85][86] Therefore, the Cu II -dpa-derivative was employed to form the SAM for oxyanion detection in aqueous media (Figure 7a).The selectivity of the OFET with the extended-gate electrode was carried out against 13 oxyanions containing phosphates, carboxylates, and nucleotides (Figure 7b).The OFET-based oxyanion sensor showed the highest response to Pi at a low concentration (<100 μM).In contrast, the sensor exhibited higher responses to diphosphates (i.e., pyrophosphate and adenosine diphosphate (ADP)) than that to Pi at a high concentration (500 μM) (Figure 7b).The observed unique transistor responses originating from analyte structure dependency and concentration dependency implied the potential of the combination of the OFET and supramolecular receptors for pattern recognition.Therefore, a qualitative assay for the simultaneous discrimination of 13 oxyanions (i.e., Pi, PPi, adenosine monophosphate (AMP), ADP, adenosine triphosphate (ATP), terephthalate, phthalate, isophthalate (IPA), malonate (Mal), oxalate (Oxa), LA, benzoate (BzO À ), and acetate (AcO À )) was performed using the OFET-based chemical sensor (Figure 7c).In the evaluation of ).Reproduced with permission. [65]opyright 2011, American Chemical Society.
the discriminatory power of the OFET-based chemical sensor, linear discriminant analysis (LDA) was employed for data processing. [57,58]The multidimensional inset data was constructed by transistor characteristics (V DS -I DS ) by adding 13 oxyanions.With the supervised method, LDA, the classification of chemical species can be achieved based on their similarity of chemical information along with a decrease in the dimension of the inset data.Figure 7c indicates the distribution of 14 clusters derived from 13 oxyanions and control with 100% correct accuracy.Notably, the classification result by LDA reflected a similarity of molecular geometries, whereby a structural tendency such as mono-, di-, and trianions was observed as a distribution of clusters.[89] In contrast, the designed extended-gate-type OFET successfully discriminated oxyanions with structural diversity by using a single sensor device.
Furthermore, the detectability of the OFET-based chemical sensor was examined by the quantification of Pi in a diluted human blood serum sample (Figure 7d).In a regression analysis, a support vector machine (SVM) was performed to evaluate the applicability to the real sample analysis. [90]The SVM is classified into machine learning methods to construct a calibration line from complicated responses.By employing the model established using the SVM, unknown concentrations of Pi at 49 and 89 μM (pink-colored circles) were simultaneously predicted in the human serum sample.
To further expand the ability of supramolecular receptor-based SAMs, a competitive interaction-driven chemical sensing for pattern recognition was proposed as the next example. [68]The target polyamines numerously exist in foods, which could be used to evaluate the process of food spoilage. [68,91,92]In addition, polyamine levels associate with cancer and age-associated diseases. [93]Hence, high-throughput analysis of polyamines is required for monitoring food quality and diagnosis.In the design of a pattern-recognition-driven chemical sensor, a competitive assay among 2-carboxymethylthio-5-mercapto-1,3,4-thiadiazole (tmt), a Cu 2þ ion, and polyamines were employed to easily obtain an information-rich inset data (Figure 8a).Competitive interactions based on the difference in binding affinities among analytes and sensor components have been used as detection principles in applications for environmental assessment, [94][95][96] diagnosis, [97,98] food chemistry, [99] etc.In this assay, the tmt derivative was modified as an SAM-based scaffold to capture the Cu 2þ ion on the extended-gate electrode.The complexation between tmt and the Cu 2þ ion induces changes in transistor characteristics, while further transistor responses are obtained by adding polyamines because of the coordination of polyamines and the Cu 2þ ion on the SAM-based scaffold.More importantly, transistor characteristics in polyamine sensing depend on the 2À (Pi) in a human blood serum sample.Reproduced with permission. [67]Copyright 2022, American Chemical Society.
difference in binding affinities between the Cu II -tmt complex and the Cu II -polyamine complex.In other words, the competitive manner contributes to obtaining various responses for pattern-recognition-driven chemical sensing.The selectivity of the OFET functionalized with the Cu II -tmt complex was investigated against 10 amine derivatives (i.e., methylamine, spermine, spermidine, trimethylenediamine, ethylenediamine, histamine, tryptamine, putrescine, cadaverine, and 1,6-diaminohexane) (Figure 8b).As shown in Figure 8b, the profiles of titration isotherms by adding 10 amines at different concentrations showed a concentration-dependent cross-reactivity, suggesting the feasibility of the OFET-based chemical sensor for simultaneous recognition of various amines.Remarkably, the response pattern constructed by the changes in transistor characteristics (i.e., (V TH -V TH0 )/V TH0 ) includes both positive and negative V TH shifts, which was attributed to the difference in the binding stoichiometry and binding affinities in the competitive recognition fashion among tmt, the Cu 2þ ion, and the target amines.The positive V TH shifts could be derived from the removal of the Cu 2þ ion upon the addition of the target amines (i.e., ethylenediamine, spermine, spermidine, trimethylenediamine, histamine, and tryptamine).In contrast, the negative V TH shifts were probably due to the capturing of the target amines (i.e., 1,6diaminohexane, cadaverine, and molecular putrescine) on the Cu II -tmt complex-attached extended-gate electrode.In addition, a response to methylamine was not observed owing to the less binding site.Therefore, the competitive assay using the OFET-based chemical sensor was applied to the qualitative and quantitative detection of various polyamines.Figure 8c exhibited a distribution of all clusters by LDA, which indicates a 100% correct classification of nine amines and a control.More importantly, positive and negative V TH shifts are contained in the inset data associated with the position of each cluster.Furthermore, the distribution of histamine and tryptamine along with close distance implied the discriminatory ability of the OFET-based sensor for the categorization of aromatic and alkyl amines.Furthermore, the OFET-based chemical sensor performed a regression analysis for the quantification of spermine levels in a diluted commercial mango juice sample to evaluate its applicability for food analysis.The unknown spermine levels in the mango juice were successfully predicted by using SVM.
Overall, this section revealed the usability of the sensing system combined with the cross-reactivity of supramolecular receptors and appropriate machine-learning methods in practical analysis situations.In chemical sensing based on competitive assays using supramolecular receptors, various responses upon analyte detection can contribute to pattern recognition, whereas the sensing mechanism requires the addition of competitors (i.e., metal ions) at each measurement for repeatable detection.Therefore, a simpler strategy for reusable chemical sensing is desirable.

OFET-Based Sensors Functionalized with MIPs for Selective Detection
As described in Section 4, the cross-reactivity of supramolecular receptors can be utilized for sensing platforms combined with powerful data-processing techniques.Meanwhile, quantum chemical calculation methods can be used for receptor designs considering the molecular geometry of analyte structures.Molecular imprinting is a method to obtain a 3D recognition network inside cavities against a specific analyte.MIPs are referred to as plastic antibodies because of their high binding affinity to analytes, which can be obtained by the following process; preorganization of a template (i.e., analyte) and building blocks (i.e., functional monomers) based on spontaneous self-assembly, polymerization, and extraction of the template from the polymerized product (Figure 9). [100,101]Therefore, optimal molecularrecognition cavities for analytes can be designed by the selection of appropriate building blocks for sensing purposes.In the fabrication of MIPs, the molar ratios of the templates and building blocks significantly affect sensor responses. [102]Polyamine] = 150 μM.Reproduced with permission. [68]Copyright 2021, The Royal Society of Chemistry.
Hence, this section shows a methodology for selective recognition using OFET-based chemical sensors functionalized with MIPs on the extended-gate electrodes.
The first example of Section 5 is an MIP-attached OFET (MIP-OFET) for the selective detection of cortisol in human saliva.Salivary cortisol referred to as a stress hormone has been used as a reliable biomarker to evaluate a state of psychological disorders.Given that accurate detection of salivary cortisol is necessary for the precise analysis of psychological disorders at nM levels, [103] various cortisol sensors have been reported.For example, supramolecular receptors based on macrocyclic structures such as cucurbit[n]urils [104] and cyclodextrins [105][106][107] have been applied to cortisol detection, whereas the selective detection of steroid hormones in host-guest chemistry is challenging because of the difficulty of designs of supramolecular receptors considering steric hindrance.Thus, the molecular imprinting method was employed to easily obtain an optimal 3D recognition network with geometric configuration against cortisol. [106,108,109],2-Diaminobenzene (1,2-DAB) was used as the monomer unit for the formation of MIP, which captures cortisol through multiple hydrogen bonds.[110,111] In addition, the monomer can be polymerized on the electrodes through an electrochemical method (i.e., cyclic voltammetry), which contributes to obtaining highly reproducible MIP layers.An appropriate molar ratio between the template (i.e., cortisol) and the functional monomer (i.e., 1,2-DAB) was determined using calculation methods (e.g., density-functional theory (DFT)) to obtain a pre-organized structure consisting of cortisol and 1,2-DAB for MIP fabrication.In the DFT calculations, monomer ratios from 1:1 to 1:5 (= cortisol:1,2-DAB) showed a drastic decrease in binding energy.In this regard, the significant changes in binding energy between the complex at the 1:5 and 1:6 ratios (%7.9%) were not observed in comparison with the change between the complex at the 1:4 and 1:5 ratios (%45%).In contrast, results of an electrochemical measurement (i.e., differential pulse voltammetry (DPV)) clarified a higher response at the 1:5 ratio to cortisol than that at the 1:6 ratio under the same sensing conditions, indicating that the monomer at a too high ratio caused a low density of the cavity in the MIP for cortisol.Thus, polymerization conditions were decided to a 1:5 ratio (Figure 10a). The MIP-OET showed quantitative shifts upon adding cortisol at different concentrations, and the LoD was determined to be 0.72 μg L À1 (i.e., 2.0 nM) based on the 3σ method (Figure 10b).Notably, the response concentrations of the MIP-OFET to cortisol were lower concentrations than that of DPV measurements, which was derived from the inherent amplification ability of the OFET.Next, a selectivity test was performed against similar structural steroid hormones (i.e., progesterone and prednisolone) and interferents in human saliva (i.e., uric acid, creatinine, glucose, and LA).As shown in Figure 10c, the MIP-OFET exhibited the highest response to cortisol, whereas weak responses to the other analytes were observed.Remarkably, the MIP-OFET showed the capability for the discrimination of the target cortisol against similar structural steroids (i.e., prednisolone).In addition, the discriminatory ability of the MIP-OFET against cortisol was evaluated in a pooled sample solution containing all the mentioned interferents, which showed a similar transistor response to cortisol in the absence of interferents (Figure 10c).Furthermore, a spike-and-recovery test against salivary cortisol was demonstrated using the MIP-OFET.The predicted datasets (red circles) of spiked cortisol in human saliva were accurately distributed on the established calibration line, which indicated 102%-110% recovery rates (Figure 10d).The accurate recovery rates suggested the reliability of the optimized MIP and usability of the MIP-OFET for the diagnosis of cortisol levels in human saliva.
The next example in Section 5 is an MIP-OFET for the selective detection of taurine (Tau), which plays an important role in organ functions. [29]In the design of MIP for Tau, dopamine and 1,2-DAB were employed as functional monomers from the perspective of not only interaction with the target Tau but also applicability to electrochemical deposition in MIP fabrication.The selected monomers form hydrogen bonds with the target Tau in a pre-organized structure based on molecular selfassembly. [112]The pre-organized structure consisting of dopamine, 1,2-DAB, and Tau was optimized by the DFT calculations at a 3 : 1 : 1 molar ratio.Figure 11a shows the isosurface figure of the Independent Gradient Model [113] of the preorganized structure, which visualized non-covalent interactions (i.e., hydrogen bonds) between the template Tau and the functional monomers.The MIP-OFET showed quantitatively negative shifts of transfer characteristics with an increase in Tau concentration, suggesting the encapsulation of Tau in the cavity of MIP.The LoD value was estimated to be 0.33 μM based on the 3σ method. [55]Remarkably, the sensitivity of the MIP-OFET was higher than that of the conventional electrochemical method (i.e., DPV), suggesting that an amplification ability of the OFET contributed to enhancing sensor sensitivity. [102]In addition, the selectivity test against similar structural analogs (i.e., alanine (Ala), arginine (Arg), aspartic acid (Asp), cysteine (Cys), glutamic acid (Glu), and malonic acid (Mal)) revealed a high discriminatory power of the MIP-OFET (Figure 11b).Moreover, the repeatability of the MIP-OFET was evaluated by alternative adding Tau and washing process.Figure 11c showed transistor responses corresponding to the capture and release of Tau.Furthermore, Tau levels in a diluted commercially available energy drink were quantified using the MIP-OFET, which revealed the potential for real-sample analysis.
The technology of molecular imprinting was further applied for the synthesis of recognition cavities to selectively detect  )).The pooled sample means the buffer solution containing all the aforementioned interferents.d) Spike-and-recovery test for cortisol in the human saliva sample.Reproduced with permission. [102]Copyright 2023, Elsevier.[Analyte] = 10 μM.c) Reusability test using the MIP-OFET for Tau.Each OFET response was evaluated by adding Tau (10 μM) and washing with an acidic solution alternatively.Reproduced with permission. [29]opyright 2021, The Royal Society of Chemistry.alkaloid drugs. [30]The analysis of alkaloid drugs is important because of their side effects causing psychological disease.Among them, atropine is a muscarinic-blocking drug used for treating intoxications by organophosphorus analogous, which causes significant side effects (e.g., hallucination, memory impairment, and blurred vision). [114]The alkaloid drug, atropine is used as a racemic structure in practical situations, while the pharmacological activity is induced by only (S)-hyoscyamine. [114]n the fields of host-guest chemistry, discrimination of chiral compounds is challenging owing to the difficulty in sophisticated receptor designs taking complementarity into account. [88,115,116]herefore, the techniques of molecular imprinting were applied to the design and synthesis of the chiral-recognition scaffold in this demonstration.MIP for selective detection of (S)-hyoscyamine was fabricated using two functional monomers (i.e., N-isopropylacrylamide (NIPAM) [117] and 2,2-dimethyl-4-pentenoic acid (DMPA)).Branch methyl groups of DMPA were expected to provide a steric effect in a molecular-recognition scaffold to discriminate the bulky structural drug. [118]A pre-organized structure consisting of NIPAM, DMPA, and (S)-hyoscyamine was optimized under the conditions at a 4:1:1 ratio (Figure 12a).The DFT results indicated a high complementarity between the template and functional monomers through six hydrogen bonds.Figure 12b clarifies the high selectivity of the MIP-OFET to (S)-hyoscyamine over muscarinic-blocking drugs (i.e., scopolamine, tropicamide, solifenacin, and oxybutynin) [114] and metabolites of atropine (i.e., tropic acid and tropine). [119]Meanwhile, the moderate responses using the MIP to tropine implied a higher contribution of the skeleton of 8-azabicyclo[3.2.1]octane than 3-hydroxy-2-phenylpropanoate in the recognition of (S)-hyoscyamine. [119]Remarkably, the MIP-OFET discriminated against very similar chemical structures between (S)-hyoscyamine and scopolamine based on a slight difference in the structure derived from an intramolecular hydrogen bond (Figure 12b). [120]Among chiral drugs, the difference in the pharmacological activity derived from chirality causes serious side effects, indicating that the development of chiral sensors allowing easy-to-determine enantiomeric excess (ee) is in high demand. [121]Therefore, the usability of the MIP-OFET for overthe-counter (OTC) drug test was investigated by the % ee determination of (S)-hyoscyamine in atropine as a racemic structure at a range from 0.2% ee to 90.2% ee. Figure 12c represents a % ee-dependent linear line associating with the shifts of V TH , which clarified the feasibility of MIP-OFETs as OTC sensors.
As described earlier, the combination of the optimized MIP designs for specific analytes and the extended-gate-type OFET showed high discriminatory abilities in real samples.Although further improvements to unify the orientation of templates in MIP fabrication processes [122] are required, the variation of MIP designs can accelerate the development of chemical sensors for real sample analysis.

Conclusion and Perspective
OFETs are representative amplification devices showing switching profiles by applying a gate voltage, indicating the potential as chemical sensor devices in combination with appropriate molecular-recognition materials.However, the applications of the OFETs in chemical sensing have been limited because of the instability of organic semiconductive materials under ambient conditions and the difficulty of the designs of molecularrecognition materials.To maximize the potential of the OFETs as chemical sensor platforms, this review described insights into the development of OFET-based chemical sensors [Analyte] = 5.0 μM.c) The threshold voltage (V TH ) corresponding to the % ee value of (S)-hyoscyamine.The overall hyoscyamine concentration was set to 5.0 μM.Reproduced with permission. [30]Copyright 2022, The Royal Society of Chemistry.
based on fusion technologies between organic electronics and molecular recognition.The employed extended-gate-type structure allows stable device operation in analyte detection owing to the isolated structure from the sensing portion.In Chapter 2, the detection principle of the extended-gate-type OFET and characterization methods for extended-gate electrodes were explained.Thus, this review summarized methodologies for the realization of bio/chemical sensors from the perspective of material designs through Chapters 3, 4, and 5.In the fabrication of biosensors consisting of enzymes or antibodies, the establishment of recognition portions considering an electron relay and orientation of charges is required to obtain reproducible sensing results.Chapter 3 focused on approaches to immobilizing biological materials using SAM-based scaffolds on the extended-gate electrodes.SAM-linked mediators constructed on the extended-gate electrodes provided uniform scaffolds for enzyme layers.Such approaches have been applied to detect biomarkers in human saliva and urine samples, which revealed high detection accuracy in real-sample analysis.The same strategy was also expanded to immunoassay for the quantification of a salivary hormone.Such biosensors based on the extended-gate-type OFET showed favorable detectability allowing highly selective and sensitive detection of specific analytes in human samples, whereas the instability of biological materials contains issues in practical sensing situations.Hence, Chapters 4 and 5 introduced supramolecular receptors and MIPs as artificial molecular-recognition materials.The inherent cross-reactivity endowed into supramolecular receptors could be used for pattern recognition.In other words, the combination of extendedgate-type OFETs with supramolecular receptors and computational data processing methods are promising sensor systems for high-throughput sensing.In Chapter 4, the representative examples of pattern-recognition-driven chemical sensing using coordination-based SAMs were summarized.Through this demonstration, the strategy for pattern recognition using a single chemical sensor device was proposed.Moreover, Chapter 5 explained the cavity designs to selectively capture specific analytes based on molecular geometry.Molecular imprinting techniques have been used to obtain synthetic molecular-recognition scaffolds, whereas the determination of molar ratios between templates and functional groups was not established.Meanwhile, this review focused on quantum chemical calculation methods to optimize pre-organized structures consisting of analytes and functional monomers, which provided optimal molar ratios for MIP fabrication.Indeed, the versatility of MIP designs was revealed in the selective detection of a steroid, a small-sized amino acid derivative, and a chiral OTC drug.Therefore, the feasibility of the extended-gate-type OFET sensors and usability for various sensing purposes were clarified by the high detection ability in different real samples toward food analysis, diagnosis, and environmental assessments.More importantly, in contrast to conventional chemical sensors based on quartz crystal microbalance and surface plasmon resonance, the extended-gate OFETs can apply to the detection of small inorganic ions [123,124] and low-molecular-weight targets. [29]In addition, the amplification ability of OFET devices over conventional electrochemical methods such as DPV in chemical sensing has been clarified through our demonstrations. [30,102]e applicability of OFETs for chemical sensing has been revealed by the uniform and reproducible switching profiles for reliable analyte detection, and robustness in both the DC bias stress and the cycle tests. [125]Meanwhile, the employment of printing fabrication methods such as ink-jet [43,126] and gravure printing [127] instead of drop-casting methods in device manufacturing processes could further improve the transistor characteristics for chemical sensing. [125]In addition, the necessity of a reference electrode in the extended-gate-type OFETs could be a bottleneck in on-site analysis.[130][131][132] Moreover, the applicability of the extended-gate-type OFETs for flexible devices depends on differences in organic semiconductive materials.For example, higher mechanical stability was observed in an OFET made of a polymer semiconductive layer than in a small molecular semiconductor. [133]Thus, the types of semiconductive materials affect not only sensing ability but also sensing applications.The authors believe that the proposed strategy based on the interdisciplinary field of organic electronics and molecular-recognition chemistry could accelerate the establishment of chemical sensor devices for real-sample analysis.

Figure 1 .
Figure 1.a) Schematic illustration of an extended-gate-type organic field-effect transistor (OFET)-based chemical sensor and chemical structures of the representative organic semiconductive materials.b) Conceptual figure illustrating changes in transistor characteristics by analyte capture.The terms V TH and Q indicate threshold voltage and charge density, respectively.

Figure 3 .
Figure 3. a) Schematic illustration of the laccase-linked mediator immobilized on the extended-gate Au electrode for the selective detection of urinary dopamine.b) Time-dependent change in drain current (I DS ) of the OFET-based enzymatic sensor with an increase in dopamine concentrations.[Dopamine] = 0-1000 μM.The inset shows the relationship between the time-dependent change in I DS and adding dopamine concentrations.c) Selectivity test results ((V TH -V TH0 )/V TH0 ) against five analytes.[Analyte] = 50 μM.d) Spike-and-recovery test for dopamine in the human urine sample.Reproduced with permission.[38]Copyright 2023, Elsevier.

Figure 2 . 3 À.
Figure 2. a) Schematic illustration of the nitrate reductase-linked mediator immobilized on the extended-gate Au electrode for the selective detection of salivary NO 3 À .The nitrate reductase is activated in the presence of Na 2 S 2 O 4 .b) Selectivity test results against five anions.[Anion] = 0-60 μM.The inset represents the changes in the threshold voltages ((V TH -V TH0 )/V TH0 ) by adding NO 3 À at low concentrations.c) Correlation between changes in V TH of the OFET and NO 3 À levels in diluted human saliva (human saliva: HEPES = 1:9, v/v) with Na 2 S 2 O 4 .[NO 3À ] = 24-54 μM.Reproduced with permission.[33]

Figure 4 .
Figure 4. a) Nonspecific binding model: analytes are physically adsorbed on the untreated electrode.Analyte capturing models on the electrode functionalized b) with antibodies and c) self-assembled monolayer (SAM)-linked antibodies.

Figure 6 .
Figure 6.a) Library of recognition groups for a chemiresistor array and target volatile organic compounds (VOCs).b) The fingerprint-like response pattern of the conductive device to 20 types of VOCs at 1% of their saturated vapor pressures (n = 3).c) Principal component score plots of the conductive array comprising eight multi-walled carbon nanotube resistance sensors for 20 types of VOCs (n = 3).Reproduced with permission.[65]Copyright 2011, American Chemical Society.

Figure 11 .
Figure 11.a) Isosurface figure of the Independent Gradient Model of the pre-optimized self-assembled structure comprising dopamine, 1,2-DAB, and Tau at a 3:1:1 molar ratio.b) Selectivity test result against seven analytes.[Analyte]= 10 μM.c) Reusability test using the MIP-OFET for Tau.Each OFET response was evaluated by adding Tau (10 μM) and washing with an acidic solution alternatively.Reproduced with permission.[29]Copyright 2021, The Royal Society of Chemistry.

Yui
Sasaki received her Ph.D. degree from the University of Tokyo in 2020 under the supervision of Associate Professor Tsuyoshi Minami.Between 2018 and 2021, she was a JSPS Research Fellow for Young Scientists (DC1) and JSPS postdoctoral fellow (PD) at the University of Tokyo.During her Ph.D. course, she worked at East China Normal University and University of Technology of Compiègne as a collaborative researcher.She worked as a project researcher between 2021 and 2022 and was promoted to a project research associate at the same university.She has also been working on a project of Nano Materials for New Principle Devices, as a JST PRESTO researcher since 2023.Her research interests include bio/chemical sensing based on supramolecular chemistry.Tsuyoshi Minami obtained his Ph.D. from Tokyo Metropolitan University in 2011.During his Ph.D. research, he worked at University of Bath as a collaborative researcher.Between 2011 and 2013, he worked as a postdoctoral research associate at Bowling Green State University and was appointed as a research assistant professor in 2013.In 2014, he proceeded to Yamagata University as an assistant professor.Thereafter, he was appointed as a lecturer at the University of Tokyo in 2016, and associate professor since 2019.His research interests are supramolecular analytical chemistry, self-assembled materials, nanoparticles, and organic transistors for bio/chemical-sensing applications.

Table 1 .
Summarized examples of extended-gate-type OFET-based chemical sensors.